# Is AI progress slowing down?

Markdown copy of https://realalpha.app/q/is-ai-progress-slowing/ on Real Alpha, the best minutes from long interviews with the people building AI. Cite and link the page, not this copy.

[All questions](https://realalpha.app/q/)

Question

Of the 5 moments we cut that answer this, 3 say it’s shifting to new methods and 2 say progress is slowing. Misha Laskin said on No Priors (Oct 9, 2026) that progress is shifting toward reinforcement learning and AI-assisted research as ways to improve efficiency. Demis Hassabis said on 20VC (Apr 7, 2026) that scaling still brings substantial gains, though the returns are smaller than they were at the start.

## Where they stand

- Slowing down 2
  - Dylan Field
  - Demis Hassabis
- Shifting to new methods 3
  - Misha Laskin
  - Eric Schmidt
  - Demis Hassabis

## Shifting to new methods 3

1. No Priors · Oct 9, 2026 · 4 min
   **[Frontier training costs rise 4x per generation, and AI speeds the work](https://realalpha.app/s/frontier-training-costs-rise-4x-per-generation-and-ai-speeds/)**
   > “You can kind of think about it as for every generation of model, there's a 4X multiplier in compute, roughly.”
   [Misha Laskin](https://realalpha.app/builders/misha-laskin/) Co-Founder & CEO, ReflectionAI [9:24](https://realalpha.app/v/up4sG9RM20M/?t=564)
2. Inside Blackstone · Sep 29, 2026 · 4 min
   **[Long reasoning is advancing, but original discovery remains hard](https://realalpha.app/s/long-reasoning-is-advancing-but-original-discovery-remains/)**
   > “The biggest and most important thing that has occurred in the last year is the arrival of long reasoning, deep reasoning.”
   [Eric Schmidt](https://realalpha.app/people/eric-schmidt/) CEO, Relativity Space [16:43](https://realalpha.app/v/PSbUom-p1hU/?t=1003)
3. 20VC · Apr 7, 2026 · 3 min
   **[Algorithmic breakthroughs could widen the gap between AI labs](https://realalpha.app/s/algorithmic-breakthroughs-could-widen-the-gap-between-ai/)**
   > “invent new algorithmic ideas are going to start having bigger advantage over the next few years”
   [Demis Hassabis](https://realalpha.app/builders/demis-hassabis/) Co-Founder & CEO, Google DeepMind [9:35](https://realalpha.app/v/SSya123u9Yk/?t=575)

## Slowing down 2

1. All-In · Oct 19, 2026 · 5 min
   **[Creative AI progresses slowly because its results are hard to verify](https://realalpha.app/s/creative-ai-progresses-slowly-because-its-results-are-hard/)**
   > “the fact that these are non-verifiable domains has something to do with it.”
   [Dylan Field](https://realalpha.app/people/dylan-field/) Co-founder & CEO, Figma [33:35](https://realalpha.app/v/2Zj6uMv3kTA/?t=2015)
2. 20VC · Apr 7, 2026 · 4 min
   **[AI still lacks continual learning despite strong scaling returns](https://realalpha.app/s/ai-still-lacks-continual-learning-despite-strong-scaling/)**
   > “These systems don't learn after you finish training them, after you put them out into the world.”
   [Demis Hassabis](https://realalpha.app/builders/demis-hassabis/) Co-Founder & CEO, Google DeepMind [5:06](https://realalpha.app/v/SSya123u9Yk/?t=306)

Each moment is a few minutes of a long interview, cut by Real Alpha and filed under how it answers by a model that read its claim and quote. Open a moment for its full transcript. Last moment added Oct 23, 2026.